Author: Dominik Zagórnik
The motivation behind this project was to optimize data collection from the environment with fiducial markers. The station consists of two main parts: a linearly moving carriage with an adapter for mounting sensors and a rotating profile for mounting markers. It is controlled by a ROS wrapper that sends G-code commands to a Melzi board (a basic Ender 3 3D printer motherboard) running modified Marlin firmware. The station server manages all sensors and the station itself, while the client records ROS topics into a bag file. These nodes can be run on separate devices. A typical station measurement scenario involves collecting x frames from the camera and y from the LiDAR at a given distance and marker rotation range in a stepwise manner. The acquired data can be used for further research, processed using ROS packages, or exported to a more general format. This solution was designed in short time and to be cost-effective, and it succeeded quite well.
-
serial package:
sudo apt install ros-noetic-serial
Then, create a Catkin workspace and clone the repository:
mkdir -p ~/station_ws/src
cd ~/station_Ws/src
git clone https://github.com/Escargot05/measurement-station-roscd ~/station_ws
catkin_make
echo "source ~/station_ws/devel/setup.bash" >> ~/.bashrc
source ./deve/setup.bashroscd measurement_station
./scripts/create_udev_rules.shRemember to also install the udev rules for rplidar and the astra camera.
Edit the rplidar launch file to set the serial_port parameter.
<param name="serial_port" "type="string" value="/dev/rplidar"/>-
Set the path and name for the rosbag in
station_client.launch. -
Configure the measurement scenario in
station.yaml. -
Launch the station server and client:
roslaunch measurement_station station_server.launch roslaunch measurement_Station station_cleint.launch
e- Enable station motorsd- Disable station motorsl- Rotate marker counterclockwiseh- Rotate marker clockwises- Start data collection and open the bagp- Finish data collection, close the rosbag, and stop the server
For images and simple numeric messages, rqt_bag is sufficient. However, for scans and point clouds, use rviz.
rqt_bag <path_to_rosbag>